Apertus-8B-Instruct-heretic

A decensored variant of swiss-ai/Apertus-8B-Instruct-2509, produced with Heretic v1.4.0 (directional ablation / "abliteration"). Apertus is the fully-open Swiss model - open weights, open data, 1000+ languages, trained from scratch on 15T tokens with the xIELU activation; refusal behaviour is suppressed here via targeted weight edits to the attention output and MLP down-projections rather than fine-tuning, so the multilingual instruction tuning is left largely intact.

Who this is for: developers who want an uncensored fully-open multilingual model - refusals drop from 100/100 to 9/100, with 1000+ supported languages and 64K context. Note the base model is access-gated behind an acceptable-use policy; this decensored variant keeps the Apache-2.0 license with no gate. At 8B it runs on a single consumer GPU via GGUF.

Runs on your gaming PC

Full GGUF ladder included - pick the quant that fits your card:

Weights only, at this model's native 8.1B size; add ~1 GB per 32K of context. OOM? Drop one quant level. Headroom to spare? Go one up.

Abliteration parameters

Trial 173 of a 200-trial Heretic run (seed 3198700791). direction_index was selected per layer.

Performance

Refusals on the harmful evaluation set drop from 100/100 to 9/100 - the base model refused everything tested, and all but nine of those are gone. KL divergence of 0.0643 is a moderate-fidelity edit, so expect slightly more drift in tone and formatting than a low-KL model in this collection; the multilingual instruction tuning survives intact. Note direction_index was selected per layer here rather than as a single index, which is why the edit runs wider than the single-index runs.

Why abliteration instead of fine-tuning

Fine-tuning a "helpful" persona on top of RLHF'd refusals fights the base model's training and tends to degrade coherence. Abliteration instead finds and edits the specific weight directions responsible for refusal, leaving the rest of the network (and its capabilities) untouched. See the Heretic repo and the original abliteration writeup for the mechanism.

Made with ❤️ by RACER IS OP — follow for more uncensored models

Files

Safetensors

BF16, ~8.1B parameters. The reproduce/ directory carries the full Heretic recipe - config.toml, requirements.txt, the Optuna study journal, and SHA-256 sums - so this exact model can be regenerated bit-for-bit. Reproduce it with heretic --reproduce reproduce/reproduce.json.

GGUF quantizations

Full quantization set (F16 + Q4_K_M, Q5_K_M, Q6_K, Q8_0) produced with llama.cpp.

Apertus architecture (apertus, xIELU activation) - loads natively in llama.cpp / Ollama / LM Studio / Jan.

Run llama serve -hf saidutta69/Apertus-8B-Instruct-heretic to pull the default quant.

Quickstart

# llama.cpp
llama serve -hf saidutta69/Apertus-8B-Instruct-heretic

# transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "saidutta69/Apertus-8B-Instruct-heretic"
model = AutoModelForCausalLM.from_pretrained(model_name, dtype="auto", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(model_name)

messages = [{"role": "user", "content": "Explain photosynthesis in French, then summarise the same explanation in Hindi."}]
inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=True,
                                        return_dict=True, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=1024)
print(tokenizer.decode(out[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))

Also runnable via Ollama, LM Studio, Jan, vLLM, SGLang.

Multilingual by design

Apertus natively supports 1000+ languages with long-context handling to 64K - the same prompt works across high- and low-resource languages without a separate multilingual adapter. The heretic edit only touches refusal directions, so language coverage is unchanged.

Model details

Responsible use

Refusal suppression is deliberate and works as intended: this model will comply with requests the base model would refuse, including some it shouldn't. There is no safety filtering layered on top. You are responsible for how you deploy it — don't put this behind an unmoderated public-facing endpoint serving third parties. It inherits Apertus's factual limitations and biases; abliteration removes refusal directions, it doesn't add capability or judgment.

License

Apache-2.0, inherited from the base model. The base repo ships no LICENSE file, so this card links the canonical Apache-2.0 text rather than a repo blob.

Related

• swiss-ai/Apertus-8B-Instruct-2509 — the base model
• swiss-ai/Apertus-8B-2509 — the upstream pretrained checkpoint
• RACER IS OP — Heretic Models — full collection